Introducing Netryx Astra V2: an open source engine that pinpoints where any photo was taken down to its exact coordinates. No metadata, no LLMs, rebuilt from scratch.

Netryx has released Astra V2, a completely rebuilt, open-source engine that solves the complex problem of image geolocation using only visual data. By analyzing architectural details, surface textures, and spatial patterns within a photo's pixels, the system matches these features against a database of indexed street-level imagery to return precise GPS coordinates, often accurate to within a few meters. This approach bypasses the need for embedded GPS metadata or large language models, offering a purely computer vision-based solution. The tool is robust enough to handle challenging inputs like cropped, zoomed, or partial-view images, such as a single doorway or sidewalk.

The project is openly available on GitHub, positioning itself as an accessible alternative to proprietary, paywalled geolocation services. Its core innovation is a community-driven indexing model: users can map a city once, upload the index, and others can download it in minutes, collectively building global coverage. This solves the significant computational burden of repeatedly indexing the same areas. While current limitations include dependency on pre-indexed locations, lack of real-time processing, and the need for a decent GPU, Astra V2 provides a powerful, free tool for professionals in journalism, research, and analysis who need to verify locations.